China’s approval of the national security law signals the premature end to Hong Kong’s autonomy
Jessica Jim 詹穎怡
In Bowring v HMRC  the Upper Tribunal found that a scheme designed to reduce capital gains tax due on capital payments by a trust, commonly known as the ‘flip-flop II’, was effective. This case is, of course, of interest to those who implemented flip-flop II schemes before anti-avoidance legislation was introduced to block them in the Finance Act 2003. This type of scheme is no longer effective. However, the judge’s reasoning on the meaning of indirect transfers to beneficiaries under s97(5) of the TCGA 1992 is generally applicable. The judgment therefore provides a useful insight for those involved in tax planning as to how the courts are likely to decide on similar issues in the future. However, before undertaking a more in-depth analysis of the Upper Tier’s decision, it is first useful to set out the main provisions of the legislation and the basis upon which the parties deployed their respective arguments.
Section 86 of the Taxation of Chargeable Gains Act 1992 (TCGA 1992) (known as the settlor charge) imposes a capital gains tax (CGT) charge on UK resident (or, before 6 April 2013, ordinarily resident) and domiciled settlors of offshore trusts, when the offshore trust realises a capital gain. Section 86 applies where the settlor has an interest in the settlement. The purpose of the legislation is to prevent UK individuals avoiding a charge to CGT simply by holding assets through an offshore trust (as offshore trustees are not generally liable to UK CGT). Paragraph 2(1) of Sch 5 to TCGA 1992 sets out the test for whether the settlor has an interest in the settlement. Broadly, the settlor has an interest where the trust property may be paid to (or applied for the benefit of) a defined person. A defined person includes the settlor, their spouse, and the settlor’s (and their spouse’s) issue and their spouses. Paragraph 2(1)(c) of Sch 5 to the TCGA 1992 extends this, by stating that the settlor also has an interest where any defined person enjoys a benefit directly or indirectly from any relevant property in the trust (that is, property given to the trust by the settlor).
Section 87 (the beneficiary charge) applies when s86 does not (where the settlor is either non-UK domiciled, non-UK resident, or where they do not have an interest in the trust or are deceased). Section 87 charges CGT on beneficiaries who are resident (or, before 6 April 2013, ordinarily resident) in the UK, when capital payments they receive from the trust are matched with capital gains realised by the trustees.
Section 90 ensures that the beneficiary charge continues to apply where trustees of an offshore trust transfer assets to another trust and the recipient trust makes capital payments to beneficiaries. This is achieved by treating the unmatched gains (or a pro-rated amount of the gains) of the transferor trust as having been added to the transferee trust. The transferred unmatched gains are then available for matching to distributions and benefits made by the recipient trust.
Section 97 includes a definition of ‘capital payment’ for the purposes of the beneficiary charge. Section 97(5) treats any payment made by a trustee as a capital payment that is subject to the beneficiary charge where the payment is:
Applying the general matching rules for capital payments made to UK-resident beneficiaries from offshore trusts, where the gains matched are realised in the current tax year or the year preceding the matching, the usual rate of CGT will apply. Where the gains matched are from earlier tax years, an increased tax rate will be applied, including an element of notional interest (s91, TCGA 1992). The amount of notional interest depends on how long before the matching the gains were realised. It is set at 10% of the usual CGT rate for each additional tax year that has passed since the gain was realised, up to a maximum of 60% of the usual CGT rate.
Before anti-avoidance legislation was introduced by the Finance Acts 2000 and 2003, some UK domiciled settlors of offshore trusts sought to avoid a liability to CGT under ss86 and 87, by entering into flip-flop schemes. The first version of these schemes (flip-flop I) was used where the offshore trust had assets standing at a significant but unrealised capital gain, and the trustees wanted to sell those assets and realise the gain without incurring a CGT liability for the settlor or any UK-resident and domiciled beneficiaries.
Broadly, the schemes worked as follows:
From 21 March 2000 (when Sch 4B and 4C of TCGA 1992 came into force), flip-flop I schemes no longer worked because the new legislation (introduced by Finance Act 2000) deems there to be a disposal of all trust assets where trustees of an offshore trust make either a capital distribution or a loan linked with trustee borrowing. The result is that any gains in the transferor trust are simply read through to the transferee trust (s90) or attributed under s86. The mechanism used to block flip-flop I schemes caught capital payments from both trusts, and so the anti-avoidance draftsman added an additional provision (s90(5)) which suspended the operation of s90 in the case of an inter-trust transfer to which the new anti-avoidance Sch 4B (transfers linked with trustee borrowing) applied. The reason for this suspension was that it would ensure that gains were locked in the transferor settlement so that they could then be attributed to beneficiaries under the new Sch 4B and 4C (attribution of gains to beneficiaries). This created a loophole that was quickly exploited after 21 March 2000 to create a mutation of the earlier scheme (flip-flop II).
Flip-flop II works by deliberately triggering the provisions of Sch 4B (which brings s90(5) into play so that the inter-trust transfer provisions in s90 do not apply). Starting with a trust within s87, the trustees borrow and transfer the proceeds of borrowing to a new trust in the same way as flip-flop I. As there is trustee borrowing in place at the time of the transfer, a disposal is triggered under Sch 4B triggering the latent gains in the trust. However these gains (and any existing unmatched gains) do not transfer to the new trust under s90 because s90(5) applied. This had the ultimate effect of isolating gains that could be attributed to beneficiaries under the beneficiary charge in the transferor trust and leaving clean capital in the transferee trust that can then be distributed without a CGT charge.
Mr Bowring and his sister, Miss Bowring, were UK-resident and UK-domiciled beneficiaries of an offshore trust established by their father in 1969. By 2001 the trust had unmatched stockpiled gains of about £3m. A UK-resident trust in similar terms was established in 2002, with Mr Bowring and his sister being principal beneficiaries. The trustees of the 1969 trust sold some trust assets at market value to Mr and Miss Bowring, who had borrowed funds from another family trust to pay for them. The 1969 trustees used the cash proceeds to buy gilts. They then took out a bank loan secured on the gilts and paid an amount equivalent to the loan to the 2002 trust. The trustees of the 2002 trust made distributions to Mr and Miss Bowring, but did not distribute the entire trust fund. Mr and Miss Bowring used some of the distributed capital to repay their loans. The trustees of the 1969 trust subsequently sold the gilts and repaid the bank loan, leaving a balance in the trust fund of £300. Mr Bowring paid £400,000 of the funds that he had received as a distribution from the 2002 trust to two cousins, who were also beneficiaries of both the 1969 and 2002 trusts. It had always been the intention that he would use part of the distribution to pay them with the purpose of avoiding any possibility of a charge to tax on the cousins.
HMRC issued closure notices amending the amount of tax payable by Mr and Miss Bowring through their self-assessment tax returns on the basis that they were liable to pay CGT under the beneficiary charge on the distributions received from the 2002 trust, as well as a supplemental charge under s91. Mr and Miss Bowring appealed to the First-tier Tribunal (FTT).
The tribunal judges, Barbara Mosedale and Richard Thomas, dismissed Mr and Miss Bowring’s appeals, applying the conclusions of the Special Commissioner (Sir Stephen Oliver QC) in Herman v HMRC . In Herman, although it was accepted that the flip-flop II planning was effective to prevent the realised gains in the original trust transferring into the new settlement, s97(5) was said to apply very widely so that the beneficiaries who received distributions from the transferee trust were treated as receiving capital payments indirectly from the original trust, as well as directly from the 2002 trust. They were therefore subject to tax under ss87 and 91 on the indirect capital payments.
The Bowrings appealed to the Upper Tax Tribunal (UTT).
The main issue was whether the capital payments made to both taxpayers had actually been made by the trustees of the new trust or indirectly by the 1969 trust. The UTT noted that the scheme had envisaged virtually all the transferred property being paid to the beneficiaries of the new trust and that both sets of trustees had knowingly played a part in the implementation of the scheme. However, this did not affect the fact that the original trust’s settled property had been transferred to the new trust so that when the trustees of the new trust had made the capital payments, they had done so entirely in the exercise of their own discretion, and in fact they did not distribute all of the trust fund. The UT therefore rejected HMRC’s contention that the new trust was a mere intermediary or conduit. It also found that the FTT had erred in law in holding that the capital payments had been received from the trustees of both trusts. The capital payments had been made solely by the new trust so that the purpose of the scheme was achieved.
The UT accepted that the aim of the legislation as an anti-avoidance measure had not been achieved in the present case. This was because the relevant provisions failed to transfer trust gains so that they could be matched with capital payments. However, it was not open to the UT to strain the facts to produce the outcome desired by both the legislation and HMRC.
The flop-flop II scheme has been rendered largely ineffective since the anti-avoidance legislation introduced by the Finance Act 2003 (as the gains would be matched to capital payments from either trust under the amended rules). However, the Bowring judgment nevertheless provides useful insight into what the court may consider when looking at similar planning in the future.
HMRC contended that the 2002 trust was merely a conduit for the 1969 trust. The argument was that the planning had been fixed in advance and the 2002 trust was not a real trust – it only existed to implement the planning and the trustees were never going to do anything else. One factor that the UT makes repeated reference to is that the arrangement between the two trusts in the Bowring case was not fixed and there were a number of uncertainties in the way that the scheme operated. The court was satisfied that there was no evidence to suggest that the transfer between the two trusts was made on the condition that certain future distributions would be made and the 1969 trust had no say in what the trustees of the 2002 trust chose to do. Further, no final decision was taken as to what distributions would be made and when until after the transfer had been made. Indeed, the trustees of the 2002 did, in fact, only distribute £2.4m of the £3.8m pot. This idea that the trustees of the 2002 trust acted with a considerable level of autonomy is undoubtedly one which helped secure the appellants’ success in this case. It is notable that the tribunal was most interested in whether the 2002 trust behaved as a ‘real’ trust and had the necessary power to do so. The uncontested fact that it was created as part of a flip-flop II scheme and did make considerable tax free distributions in line with this scheme was not fatal in itself.
Compare this to the recent decision in the case of Trustees of the Morrison 2002 Maintenance Trust and others v HMRC . In this case the FTT held that a carefully planned tax avoidance scheme to sell shares held in Scottish trusts and involving a number of intermediary steps formed a single composite transaction, that is, a disposal of the shares in the market by the original trusts with the chargeable gains accruing to those trusts. Applying the approach in WT Ramsay v Inland Revenue Commissioners  and later cases, Judge J Gordon Reid held that it was not a question of respecting each individual step and concluding that, as they are formally and individually immune from challenge, no fiscal liability can attach. The question was whether the relevant statutory provisions (in this case the Taxation of Chargeable Gains Act 1992), construed purposively, apply to the facts viewed realistically. He concluded that the scheme served no purpose other than avoiding a liability to tax and could therefore be analysed as an integrated whole, with the various interrelated transactions being disregarded. While the overall scheme contained a number of doubts, uncertainties and contingencies, viewed overall, it became increasingly obvious that the scheme would be carried into effect as expected, planned and intended.
A key difference between this case and Bowring is the degree of discretion in carrying out the planning. In both cases there was an overall plan including the various steps. However in Morrison, the plan was rigidly followed and there appeared to be no discretion. Contrast this with Bowring, where the trustees of the 2002 trust were aware of the plan and that the planning was essentially completed by making distributions from the 2002 trust. However it was genuinely up to them whether to make the distributions, and, as it happened, they did not distribute the whole trust fund. It is difficult to know whether the same finding would have been made if the trustees had distributed the whole trust fund, particularly if they did it very soon after the funds were received by them. It is certainly easier to demonstrate discretion when one of the parties does not follow all the steps precisely! It is probably also true that when all the steps happen as planned in a short space of time, it is easier for the tribunal to ‘step back’ and apply a Ramsey-type interpretation of what ‘really happened’.
Another point of interest is that the UTT held the capital payments were made directly by the 2002 trust. It overturned the judgment of the FTT that the payments were made by both trusts and, importantly, rejected the idea that the payment was made indirectly from the 1969 trust. This is in spite of the fact that s97 provides for the possibility that a capital payment can be made indirectly. Again it was important in this case that the 2002 was a ‘real’ trust which operated separately from the 1969 trust. That being the case, the UT held that the obvious application of the legislation to these facts is that a direct payment was made from the 2002 trust. Clearly flip-flop schemes are not the only type of double trust planning. A common theme of this type of planning is to avoid gains from one trust being matched to distributions from the other. This finding gives some comfort that it is not correct for HMRC to argue a distribution from trust 2 is an indirect payment from trust 1 and hence should be matched to trust 1 gains, even where all the assets in trust 2 originate from trust 1. It continues to be important to ensure that trust 2 is clearly a separate trust and, as seems clear from Bowring, that trust 2 has discretion and a ‘mind’ separate from trust 1.
Settled property is defined in s68 TCGA to mean any property held in trust other than as bare trustee or nominee. A trust can contain more than one fund, and it is not unusual for a single trust to be divided up in this way for family or other reasons. Where this is the case for a trust within s87, and provided no sub-fund election has been made under Sch 4ZA of the TCGA, there is a single capital gains tax pool which is available for matching to capital payments from any of the funds. As the definition of a trust is so wide, and a fund treatment would seem possible, it is a widely held concern about double trust structures that HMRC would contend that the two trusts are in fact a single settlement with two funds. On that basis the planning would be ineffective for CGT as the gains pool would be matched to distributions from trust 2 – precisely what the structure is seeking to avoid. It is curious therefore that HMRC did not run this argument in Bowring. Rather, they accepted that the 2002 trust was a separate settlement for CGT purposes but argued it had acted as a conduit for the 1969 trust.
Does this mean we can now be less concerned about a challenge on that basis? It must be assumed that HMRC did not take this line of argument for a reason. Perhaps it would cause other issues or simply be outside the purpose of the legislation which has s90 to deal with trust to trust transfers and Sch 4B and 4C for flip flop scenarios. Although arguably s90 would not be redundant, the single settlement argument would likely only apply to this type of incestuous tax planning where the new trust essentially does the same ‘job’ as the first trust. It is understood that the usual precautions were taken in setting up the 2002 trust to ensure it was not simply a mirror of the 1969 trust: there were different trustees; the 2002 trust was UK resident; and the terms of the trust differed. Perhaps these were a barrier to the argument – it seems worth continuing with this approach – but it would also seem possible to argue that these factors are a tax planning fiction and the reality is that there is one trust. This would not be inconsistent with the Ramsay line of cases where a real or commercial view is taken of the facts to which the law is then applied (see above).
Of course, since 2013, we must also consider the GAAR. To go into detail on this and its application to double trust structures would require another article. We would simply ask whether a challenge under the GAAR would be a preferred line of attack by HMRC on double trust structures particularly given that, as discussed above, Bowring confirms their effectiveness for CGT planning in a number of ways. Most notable is that the UT considered that they could not rewrite the legislation where its meaning and application to the facts is clear, even when it has a tax result outside the intention of parliament. Could the GAAR therefore be used to address this for HMRC?
We will end this article by noting that the position on the taxation of trusts with non-UK domiciled beneficiaries or settlor(s) may well change when the proposed new deemed domiciled rules come in from 6 April 2017. It is unclear exactly what the new rules will be at the time of writing.
Clive and Juliet Bowring v HMRC  UKUT 550 (TCC)
Herman & anor v HMRCC  WTLR 1201 SpC
Trustees of the Morrison 2002 Maintenance Trust & ors v HMRC  UKFTT 250 (TC)
WT Ramsay v Inland Revenue Commissioners  UKHL 1
This article first appeared in the Trusts and Estates Law and Tax Journal.
For further information on any of the issues raised, please contact a member of our private client team.
Skip to content Home About Us Insights Services Contact Accessibility